Friday, 14 October 2011

The environmental cost of science

I am now in Munich, for the conference Deep Metazoan Phylogeny 2011. A lot of interesting talks and posters about phylogenies of various animals like mnemipopsis, rhabdura, nemertean, remipedia or buddenbrockia  (and a lot of interesting Bavarian beers too). What I can say is that a taxonomy is not always fixed but can vary among data and methods. Among these talks, one attracted my attention, not only on the scientific content, but on the ecological message. Hervé Philippe, professor from Montreal specialised in phylogeny (ie the CAT model), has presented in 2-3 slides the environmental impact of building huge phylogeny. But this also applies to large-scale research, like next-generation of whole genome sequencing and large-hadron collider (LHC) for example. For that, he takes into account the cost of CO2 produce by CPU/hour. As he wrote in the acknowledgment in one of his last papers in Nature:
[…] H.P. is funded by Canada Research Chairs, Natural Sciences and Engineering Research Council and Réseau Québécois de Calcul de Haute Performance for computational resources: more than 220,000 central processing unit (CPU)–hours were used producing at least 7 tonnes of CO2 excluding grey energy. [...]
He had also noticed the impact and question the usefulness of scientific conferences in the era of Internet (Philippe, Trends in Genetics 2008).
Less is more: decreasing the number of scientific conferences to promote economic degrowth
Today’s scientists are confronted with a serious paradox: although the goal of their research is often to mitigate the negative impact of human activities (e.g. loss of biodiversity), the research community itself can be a significant contributor to the problem. Moreover, grant evaluation criteria (e.g. number of publications, number of presentations given at international conferences) strongly favour activities that have a significant impact on the environment.
He mainly suggests and/or tries to:
  • Reduce the pace of scientific research
  • Attend to maximum one international conference per year.
  • Write more reviews to reduce the duplication of already published results.
  • Work in experimental fields around the lab instead of moving to other exotic countries.
In brief, he suggests to apply the degrowth (“décroissance” in French) of science. Of course not easy to apply, but efforts could be made in this direction.

However, while I mostly agree with these points, I think it is really difficult for young scientists (like me) to reduce the pace of scientific research (ie number of computations, wet-lab experiments, world-wide field investigations). Because we are all trying to secure our position, we need to produce scientific results and to publish them, if possible in good journals (the famous “publish or perish” formula).

I have also some concerns about the estimation of CO2 usage per CPU. What about the use of super-computers in countries where electricity is mainly produced by dams (hydroelectric), like in Norway (100%) or in Switzerland (55%)?

In any cases, despite being slightly pessimistic, these remarks merit to be thought and to be more investigated.

This was my green post of the month. ;)


1 comment:

  1. Thanks for attracting our attention to this. I think that overall, increasing knowledge will have a positive impact on the environment (hey, how did we learn about global warming in the first place? and how do we predict its effects on various species?). But green computation is a growing area in high performance computing, and one that is worth paying attention to.

    As for travel: we should be careful not to overdo it, or we will have Asiatic, American, and European science not talking to each other, which would be very dangerous.